Stroke Prevention in Heart Failure and Sinus Rhythm: Where do we Go from Here?

This article refers to ‘CHA2DS2-VASc score and adverse outcomes in patients with heart failure with reduced ejection fraction and sinus rhythm’ by S. Ye et al., published in this issue on pages 1261–1266. Stroke is a clinically relevant, but infrequent adverse outcome in heart failure (HF) that may be of dire consequence in the life course of these patients.1 Given the substantial decrement in quality of life, attendant costs, and hospitalization burden referable to this morbid event,2 stroke prevention may be a worthwhile investment. Even in the absence of concomitant atrial fibrillation or flutter (AF), patients with HF who are in sinus rhythm have an increased propensity of stroke and systemic embolism (SSE) compared with the general population. The risk appears to be similar in HF patients regardless of EF.3 Mechanistically, HF augments risk of thrombosis via the Virchow's triad: ventricular stasis of blood, elevated circulating prothrombotic factors, and endothelial injury. Beyond these mechanistic contributions, HF is also strongly associated with AF, and many patients with HF may have an undiagnosed and thus untreated burden of AF that further increases this thromboembolic risk. An insufficient evidence base does not allow current US and European guidelines to recommend routine anticoagulation in patients with HF in the absence of another compelling reason (i.e. AF or prior venous thromboembolism).4, 5 In this editorial, we address open questions related to contemporary stroke prevention in HF. (i) Are there high-risk HF patients in sinus rhythm where the risk/benefit calculus favours anticoagulation? (ii) What clinical prediction tools are available to identify these enriched subsets? A number of randomized clinical trials have studied the role of routine anticoagulant and antiplatelet therapies in HF with reduced ejection fraction (HFrEF). Despite the importance of the question, a common theme has been challenges in recruitment and thus a limitation in the power to reach a definitive endpoint. Warfarin, when compared with aspirin (300–325 mg daily), in the early WASH (Warfarin/Aspirin Study in Heart Failure)6 and HELAS (Heart Failure Long-Term Antithrombotic Study)7 trials failed to demonstrate significant benefit on clinical endpoints including mortality, but did increase risk of major haemorrhage in small cohorts of HFrEF. The larger WATCH (Warfarin and Antiplatelet Therapy in Chronic Heart Failure) trial compared aspirin vs. clopidogrel vs. warfarin in 1587 patients with HFrEF.8 This ambitious trial was prematurely terminated with no observed significant difference in major adverse cardiovascular events across treatment groups over 23 months of follow-up,8 but power was a concern. The most pivotal study to date has been the WARCEF (Warfarin versus Aspirin in Reduced Cardiac Ejection Fraction) trial, which revealed that a reduction in ischaemic stroke with warfarin compared with aspirin 325 mg daily was offset by a significant increase in major haemorrhage in 2306 patients with HFrEF (≤35%) and sinus rhythm.9 There was no difference between the warfarin- and aspirin-treated patients in the primary composite endpoint of death, ischaemic stroke, or intracerebral haemorrhage.9 Yet, the results of WARCEF should be interpreted with caution as the trial did not reach target enrolment, which resulted in a power reduction from 89% to 65% despite an extended (average 3.5 years) follow-up interval. These well-intentioned but underpowered trials infer but offer no definitive proof that routine anticoagulation with a vitamin K antagonist may potentially reduce risk of ischaemic stroke at the cost of excess major bleeding without influencing mortality in patients with chronic HFrEF and sinus rhythm. This is an unsettling conclusion and does not allow for a clear clinical directive. In this issue of the journal, Ye et al.10 present a post hoc analysis of the WARCEF trial evaluating the capacity of the CHA2DS2-VASc risk score to predict death, ischaemic stroke, and major haemorrhage in 2224 patients with HF in sinus rhythm. Rates of the composite endpoint and its components were significantly higher with each 1-point increment in CHA2DS2-VASc score. Importantly, the investigators did not detect heterogeneity in treatment effect of warfarin compared with aspirin based on CHA2DS2-VASc levels for all tested clinical outcomes. This study adds to a prior observational investigation of a Danish cohort of 40 000 HF patients with and without AF, which described an association between CHA2DS2-VASc score and risk of ischaemic stroke, thromboembolism, and death.11 Interestingly, among this Danish cohort, high (≥4) CHA2DS2-VASc scores were associated with greater risk of thromboembolism in patients without AF compared with those with AF. Another large study from a Danish registry corroborated these findings.12 This observational study showed that the CHA2DS2-VASc score predicted SSE risk independent of incident AF during follow-up, and that rates of SSE were highest during the first year after HF hospitalization.12 The authors should be recognized for their pursuit of this important clinical question. The findings though provocative are not definitive. WARCEF was underpowered with a small absolute number of clinical events and few participants with high CHA2DS2-VASc scores. Most importantly, we must exercise care in interpretation of this exploratory analysis given the overall neutral WARCEF trial experience. Finally, as the CHA2DS2-VASc risk score was originally derived and validated from a cohort of AF patients, there is an inherent risk of misapplication when used in a population for which it was not originally intended. As such, these data in isolation are unable to support the use of the CHA2DS2-VASc score in guiding anticoagulation decision-making in this cohort, but do make evident the potential utility of a validated stroke prediction tool for patients with HF in sinus rhythm. Novel risk models incorporating readily available clinical parameters (e.g. NYHA functional class) and laboratory markers (e.g. natriuretic peptide levels) that are unique to the HF population have been proposed to inform risk of SSE, and offer initial promise.13 There are several techniques that have been proposed in order to enhance risk models for SSE in the HF population, which encompass improving rhythm surveillance, employing novel imaging strategies, and monitoring of circulating prothrombotic factors. These emerging modalities may allow clinicians to identify specific contributors to thromboembolic risk that are unique to the HF population. Given the well-characterized association between HF and AF, it is certainly plausible that enhanced stroke risk in HF is secondary to undiagnosed AF. This may be particularly true for patients with HF with preserved EF, in which AF co-exists in nearly two-thirds of patients.14 Further research is required to examine the utility of upfront, routine rhythm surveillance via long-term cardiac monitoring in this population. Several imaging modalities have been investigated in patients with AF to assist in ascertaining stroke risk; these imaging techniques may offer similar utility in the HF population without AF. Thrombogenesis in AF is not only secondary to absent atrial contraction, but is also due to structural changes of the atria, resulting in atrial fibrosis and diminished function, collectively known as atrial myopathy.15 Atrial myopathy can be characterized through functional imaging studies, including transoesophageal echocardiography. Several studies have shown that detection of decreased left atrial appendage flow velocity is associated with increased risk of embolic stroke.16 Interestingly, decreased flow velocities in AF patients who were in sinus rhythm at the time of the study were still associated with stroke risk. Similar results have been noted using cardiac magnetic resonance imaging.15 Patients with HF are at high risk for atrial myopathy and diminished atrial flow velocities. Further investigations are required to determine the incremental utility of these imaging tools beyond clinical risk scores in stroke risk prediction in HF and sinus rhythm. A number of prothrombotic factors are known to be elevated in patients with HF compared with the general population. In addition to increased platelet activity, fibrinogen levels, von Willebrand factor, and plasma viscosity are elevated in HF patients in sinus rhythm.17 Validation of these blood markers in broad HF patient populations will be required prior to their routine application in risk stratification. Despite major advances and step-wise declines in HF-related mortality in HFrEF using pharmacotherapies disrupting secondary maladaptive neurohormonal pathways, limited progress has been made in understanding and potentially ameliorating the heightened thrombotic state in HF. Routine use of warfarin in HFrEF and sinus rhythm appears to result in undue risk of haemorrhage when compared with aspirin. Ongoing studies utilizing non-vitamin K oral anticoagulants in the HF population offer promise, as these agents tend to have a more favourable bleeding profile than warfarin. The international COMMANDER-HF trial will examine the efficacy of low-dose rivaroxaban (2.5 mg twice daily) in reducing risk of death, stroke, and myocardial infarction in patients with HFrEF (≤40%) and CAD without AF.18 The trial is expected to randomize ∼5000 patients with recent HF hospitalization and elevated natriuretic peptides. Although a relatively small proportion of patients with HF experience stroke, it represents a potentially devastating and functionally impactful event. While rates of stroke are expected to rise with increasing age and cardiometabolic co-morbidities in the HF population, the overall clinical impact of the prothrombotic phenotype of HF in the disease's natural history is still unclear. Long-term epidemiological studies are required to ascertain whether the heightened prothrombotic state in HF truly translates to increased SSE burden, or whether these patients succumb to competing events, such as HF death, sudden death, and non-cardiovascular morbidity/mortality. At this juncture, antithrombotic therapy is not recommended when used routinely in patients with HF and sinus rhythm. More extensive rhythm surveillance (to facilitate detection of undiagnosed AF) and rigorous risk factor modification serve as important adjunctive opportunities to optimize stroke risk in the HF population. Our understanding of the pathophysiology and drivers of thrombogenesis in HF and sinus rhythm remains incomplete. Multimodality risk stratification, including clinical variables and novel imaging techniques of intracardiac haemodynamics, may not only identify patients at particularly high risk for SSE, but could also provide key insights into comprehensively defining the prothrombotic phenotype in HF. Conflict of interest: none declared.